---
title: "#rstats Twitter Explorer"
output: 
  flexdashboard::flex_dashboard:
    orientation: rows
    vertical_layout: scroll
    source_code: embed
    theme:
      version: 4
      bootswatch: yeti
    css: styles/main.css
---

```{r load_package, include=FALSE}
devtools::load_all()
```

```{r setup, include=FALSE, cache=TRUE}
library(flexdashboard)
library(rtweet)
library(dplyr)
library(stringr)
library(tidytext)
library(lubridate)
library(echarts4r)
library(DT)

rstats_tweets <- read_twitter_csv("data/rstats_tweets.csv")

count_timeseries <- rstats_tweets %>%
  ts_data(by = "hours")

tweets_week <- rstats_tweets %>%
  filter(as_datetime(created_at) %within% interval(floor_date(today(), "week"), today()))

tweets_today <- rstats_tweets %>%
  filter(created_at == today())

by_hour <- rstats_tweets %>%
  group_by(hour = hour(created_at)) %>%
  summarise(count = n()) %>%
  ungroup()

number_of_unique_tweets <- get_unique_value(rstats_tweets, text)

number_of_unique_tweets_today <-
  get_unique_value(tweets_today, text)

number_of_tweeters_today <- get_unique_value(tweets_today, user_id)

number_of_likes <- rstats_tweets %>%
  pull(favorite_count) %>%
  sum()

top_tweeters <- rstats_tweets %>%
  group_by(user_id, screen_name, profile_url, profile_image_url) %>%
  summarize(engagement = (sum(retweet_count) * 2 + sum(favorite_count)) / n()) %>%
  ungroup() %>%
  slice_max(engagement, n = 10, with_ties = FALSE)

top_tweeters_format <- top_tweeters %>% 
  mutate(
    profile_url = stringr::str_glue("https://twitter.com/{screen_name}"),
    screen_name = stringr::str_glue('@{screen_name}'),
    engagement = formattable::color_bar("#a3c1e0", formattable::proportion)(engagement)
  ) %>%
  select(screen_name, engagement)

top_hashtags <- rstats_tweets %>%
  tidyr::separate_rows(hashtags, sep = " ") %>%
  count(hashtags) %>%
  filter(!(hashtags %in% c("rstats", "RStats"))) %>%
  slice_max(n, n = 10, with_ties = FALSE) %>%
  mutate(
    number = formattable::color_bar("plum", formattable::proportion)(n),
    hashtag = stringr::str_glue(
      '#{hashtags}'
    ),
  ) %>%
  select(hashtag, number)

word_banlist <-  c("t.co", "https", "rstats")
top_words <- rstats_tweets %>%
  select(text) %>%
  unnest_tokens(word, text) %>%
  anti_join(stop_words) %>%
  filter(!(word %in% word_banlist)) %>%
  filter(nchar(word) >= 4) %>% 
  count(word, sort = TRUE) %>%
  slice_max(n, n = 10, with_ties = FALSE) %>%     mutate(number = formattable::color_bar("peachpuff", formattable::proportion)(n)) %>%
  select(word, number)

top_co_hashtags <- rstats_tweets %>% 
  unnest_tokens(bigram, hashtags, token = "ngrams", n = 2) %>% 
  tidyr::separate(bigram, c("word1", "word2"), sep = " ") %>% 
  filter(!word1 %in% c(stop_words$word, word_banlist)) %>% 
  filter(!word2 %in% c(stop_words$word, word_banlist)) %>% 
  count(word1, word2, sort = TRUE) %>% 
  filter(!is.na(word1) & !is.na(word2)) %>% 
  slice_max(n, n = 100, with_ties = FALSE)
```


Home
====

Row
-----------------------------------------------------------------------

### Tweets Today

```{r tweets_today}
valueBox(number_of_unique_tweets_today, icon = "fa-comment-alt", color = "plum")
```

### Tweeters Today

```{r tweeters_today}
valueBox(number_of_tweeters_today, icon = "fa-user", color = "peachpuff")
```

### #rstats Likes

```{r likes}
valueBox(number_of_likes, icon = "fa-heart", color = "palevioletred")
```

### #rstats Tweets

```{r unique_tweets}
valueBox(number_of_unique_tweets, icon = "fa-comments", color = "mediumorchid")
```

Row {.tabset .tabset-fade}
-----------------------------------------------------------------------

### Tweet volume

```{r tweet_volume}
this_month <- floor_date(today(), "month")

count_timeseries %>%
  e_charts(time) %>%
  e_line(n, name = "# of tweets", smooth = TRUE, legend = FALSE) %>%
  e_x_axis(
    type = "time",
    formatter = htmlwidgets::JS(
      "function(value){
        let date = new Date(value);
        
        label = `${date.getDate()}-${(parseInt(date.getMonth()) + 1)}-${date.getFullYear()}`;
        return label;
      }"
    )
  ) %>%
  e_axis_labels(y = "Tweets") %>%
  e_theme("westeros") %>%
  e_tooltip(trigger = "axis", formatter = htmlwidgets::JS("
    function(params) {
      let date = new Date(params[0].value[0])
      let options = { year: 'numeric', month: 'short', day: 'numeric', hour: 'numeric'}
      let title = `${date.toLocaleDateString('en-US', options=options)}`
      let num = `${params[0].value[1]} tweets`
      return(`${title}
${num}`);
    }")) %>% 
  e_datazoom(type = "slider") %>%
  e_zoom(
    dataZoomIndex = 0,
    start = 70,
    end = 100
  ) %>% 
  e_zoom( 
    dataZoomIndex = 0,
    startValue = today() - 7,
    endValue = today(),
    btn = "weekBtn"
  ) %>% 
  e_zoom( 
    dataZoomIndex = 0,
    startValue = this_month,
    endValue = today(),
    btn = "monthBtn"
  ) %>% 
  e_button(
    id = "weekBtn",
    position = "top",
    class = "btn btn-primary btn-sm",
    "This Week"
  ) %>% 
  e_button(
    id = "monthBtn",
    position = "top",
    class = "btn btn-primary btn-sm",
    "This Month"
  )
```

### Tweets by Hour of Day

```{r tweets_by_hour}
by_hour %>%
  e_charts(hour) %>%
  e_step(count, name = "Tweets", step = "middle", legend = FALSE) %>%
  e_x_axis(
    min = 0,
    max = 23,
  ) %>%
  e_axis_labels(x = "Time of Day (UTC)", y = "Tweets") %>%
  e_theme("westeros") %>%
  e_tooltip(trigger = "axis", formatter = htmlwidgets::JS("
    function(params) {
      let title = `${params[0].value[0]}h`
      let num = `${params[0].value[1]} tweets`
      return(`${title}
${num}`);
    }"))
```

Row
-----------------------------------------------------------------------

### 💗 Most Liked Tweet Today {.tweet-box}

```{r most_liked}
most_liked_url <- tweets_today %>%
  slice_max(favorite_count)

get_tweet_embed(most_liked_url$screen_name, most_liked_url$status_id)
```

### ✨ Most Retweeted Tweet Today {.tweet-box}

```{r most_rt}
most_retweeted <- tweets_today %>%
  slice_max(retweet_count)

get_tweet_embed(most_retweeted$screen_name, most_retweeted$status_id)
```

### 🎉 Most Recent {.tweet-box}

```{r most_recent}
most_recent <- tweets_today %>%
  slice_max(created_at, with_ties=FALSE)

get_tweet_embed(most_recent$screen_name, most_recent$status_id)
```

Rankings
=========

Row
-----------------------------------------------------------------------

### Top Tweeters

```{r top_tweeters}
top_tweeters_format %>%
  knitr::kable(
    format = "html",
    escape = FALSE,
    align = "cll",
    col.names = c("User", "Engagement/Tweet "),
    table.attr = 'class = "table"'
  )
```

Where Engagement is `RT * 2 + Favourite`

### Network of top tweeters

Relationships in the graph describe replies and quote retweets from the top tweeters
that also have the hashtag.

```{r top_tweeters_net}
edgelist <-
  network_data(rstats_tweets %>% unflatten(), "reply,quote")
nodelist <- attr(edgelist, "idsn") %>%
  bind_cols()

top_edges <- edgelist %>%
  filter((from %in% top_tweeters$user_id) |
           (to %in% top_tweeters$user_id))

top_nodes <- nodelist %>%
  filter((id %in% top_edges$from) | (id %in% top_edges$to)) %>%
  mutate(is_top = ifelse((id %in% top_tweeters$user_id), "yes", "no"),
         size = 10)

e_charts() %>%
  e_graph() %>%
  e_graph_nodes(top_nodes, id, sn, size, category = is_top, legend = FALSE) %>%
  e_graph_edges(top_edges, from, to) %>%
  e_tooltip()
```

Row
-----------------------------------------------------------------------

### Top Words

```{r top_words}
top_words %>%
  knitr::kable(
    format = "html",
    escape = FALSE,
    align = "cll",
    col.names = c("Word", "Count"),
    table.attr = 'class = "table"'
  )
```

Row
-----------------------------------------------------------------------

### Top Hashtags

```{r top_hashtags}
top_hashtags %>%
  knitr::kable(
    format = "html",
    escape = FALSE,
    align = "cll",
    col.names = c("Hashtag", "Count"),
    table.attr = 'class = "table"'
  )
```

Excluding `#rstats` and similar variations

### Common co-occuring hashtags

Hashtags that occur together, grouped by community detection

```{r co_hashtags}
top_co_hash_nodes <- tibble(
  nodes = c(top_co_hashtags$word1, top_co_hashtags$word2)
) %>% 
  distinct()

e_chart() %>% 
  e_graph() %>% 
  e_graph_nodes(top_co_hash_nodes, nodes, nodes, nodes) %>% 
  e_graph_edges(top_co_hashtags, word1, word2) %>% 
  e_modularity()
```


Data 
==============

### Tweets in the current week {.datatable-container}

```{r datatable}
tweets_week %>%
  select(
    status_url,
    created_at,
    screen_name,
    text,
    retweet_count,
    favorite_count,
    mentions_screen_name
  ) %>%
  mutate(
    status_url = stringr::str_glue("On Twitter")
  ) %>%
  datatable(
    .,
    extensions = "Buttons",
    rownames = FALSE,
    escape = FALSE,
    colnames = c("Timestamp", "User", "Tweet", "RT", "Fav", "Mentioned"),
    filter = 'top',
    options = list(
      columnDefs = list(list(
        targets = 0, searchable = FALSE
      )),
      lengthMenu = c(5, 10, 25, 50, 100),
      pageLength = 10,
      scrollY = 600,
      scroller = TRUE,
      dom = '<"d-flex justify-content-between"lBf>rtip',
      buttons = list('copy', list(
        extend = 'collection',
        buttons = c('csv', 'excel'),
        text = 'Download'
      ))
    )
  )
```